Performance Evaluation Criteria and Information Processing in Complex Decision Making

We use two experiments to examine how the design of performance evaluation criteria influences agents' information processing in complex decision-making tasks. Drawing on goal hierarchy theory, we predict that, when agents are evaluated on the actions they take to perform their task (low-level evaluation criteria), they will focus primarily on salient cues in the task setting, whereas evaluating how agents’ work impacts organizational objectives (high-level evaluation criteria) will increase their attention to less salient yet relevant information cues. As predicted, in the first experiment, we find that agents exhibit the numerosity heuristic under low-level evaluation criteria, indicative of over-attention to salient task cues, but not under high-level criteria. In the second experiment, we find that, when making performance predictions, agents under low-level evaluation criteria fixate on summary earnings measures while not attending adequately to less salient cues, and high-level evaluation criteria mitigate this fixation. Taken together, these findings provide convergent support for our theory. We discuss the implications of our findings for accounting research and practice.